Pox neuro control of cell lineages that give rise to larvalpoly-innervated external sensory organs in Drosophila

Yanrui Jiang 1, Werner Boll, Markus Noll n

Institute of Molecular Life Sciences, University of Zürich, Winterthurerstr. 190, CH-8057 Zürich, Switzerland

a r t i c l e i n f o

Article history:Received 4 July 2014Received in revised form15 October 2014Accepted 16 October 2014Available online 24 October 2014

Keywords:DrosophilaPox neuroPNSLarval p-es organsCell lineagesMitosisApoptosis

a b s t r a c t

The Pox neuro (Poxn) gene of Drosophila plays a crucial role in the development of poly-innervatedexternal sensory (p-es) organs. However, how Poxn exerts this role has remained elusive. In this study,we have analyzed the cell lineages of all larval p-es organs, namely of the kölbchen, papilla 6, and hair 3.Surprisingly, these lineages are distinct from any previously reported cell lineages of sensory organs.Unlike the well-established lineage of mono-innervated external sensory (m-es) organs and a previouslyproposed model of the p-es lineage, we demonstrate that all wild-type p-es lineages exhibit thefollowing features: the secondary precursor, pIIa, gives rise to all three support cells—socket, shaft, andsheath, whereas the other secondary precursor, pIIb, is neuronal and gives rise to all neurons. We furthershow that in one of the p-es lineages, that of papilla 6, one cell undergoes apoptosis. By contrast in Poxnnull mutants, all p-es lineages have a reduced number of cells and their pattern of cell divisions ischanged to that of an m-es organ, with the exception of a lineage in a minority of mutant kölbchen thatretains a second bipolar neuron. Indeed, the role of Poxn in p-es lineages is consistent with thespecification of the developmental potential of secondary precursors and the regulation of cell divisionbut not apoptosis.

& 2014 Elsevier Inc. All rights reserved.

Introduction

The peripheral nervous system (PNS) of Drosophila with itssensory organs provides a simple paradigm for the study of a varietyof complex processes, such as pattern formation, cell determinationand differentiation, asymmetric cell division, axonal pathfinding, orthe formation and evolutionary diversification of organs in differentsegments and species (Jan and Jan, 1993; Bellaïche and Schweisguth,2001; Betschinger and Knoblich, 2004; Lai and Orgogozo, 2004). Likeall holometabolous insects, Drosophila has a larval and an adult PNS.The former develops during embryogenesis and consists of only a fewtypes of sensory organs that are arranged in a segmentally repeated,stereotyped pattern (Hertweck, 1931; Dambly-Chaudière and Ghysen,1986; Ghysen et al., 1986; Bodmer and Jan, 1987; Hartenstein, 1988;Campos-Ortega and Hartenstein, 1997). They consist of few cells,which in most cases only contain a single neuron. They are eitherinternal, such as chordotonal organs or single multidendritic (md)neurons, or external sensory (es) organs, which are further subdividedinto mono- (m-es) or poly-innervated (p-es) es organs, depending onwhether they are innervated by a single bipolar neuron or by two orthree bipolar neurons. While most es organs are mono-innervated

and form either a dome-like campaniform sensillum (papilla) or atrichoid-type sensillum (hair), two p-es organs are present in eachthoracic and abdominal hemisegment. In the thorax, these are peg-like basiconical sensilla, named kölbchen (Hertweck, 1931), and arelocated in a ventral and dorsal position in T1, or in a ventral and lateralposition in T2 and T3. In the abdomen (A1–A7), the p-es organsconsist of a small dorsal hair, h3, and a lateral papilla, p6 (Dambly-Chaudière and Ghysen, 1986) that are homologous to those of thethorax, as evident from their transformation into kölbchen inbithoraxoid embryos (Dambly-Chaudière and Ghysen, 1987).

Each sensory organ is derived from a sensory organ precursor(SOP) cell. During early neurogenesis, groups of cells in theneuroectoderm acquire the potential to become neuronal precur-sor cells by the expression of proneural genes, such as the genesof the achaete-scute complex. From each group, a single cell isselected to become a neuronal precursor through a process called“lateral inhibition”, which is mediated by the neurogenic genes,like Notch and Delta, which silence the proneural genes in all cellsof the proneural group except the SOP. The identity of an SOP cellis further determined by the action of neuronal precursor genesand neuronal precursor-type selector genes that are required toinitiate the precise developmental program of a particular sensoryorgan (for a review, see Jan and Jan, 1993). Once an SOP cell isdetermined, it will follow a series of asymmetric cell divisions togive rise to a number of cells, which include the external sensoryorgan cells (three support cells and one or several bipolar neurons)

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/developmentalbiology

Developmental Biology

http://dx.doi.org/10.1016/j.ydbio.2014.10.0130012-1606/& 2014 Elsevier Inc. All rights reserved.

n Corresponding author. Fax: þ41 44 635 6829.E-mail address: markus.noll@imls.uzh.ch (M. Noll).1 Present address: Biozentrum, University of Basel, Klingelbergstrasse 50/70,

CH-4056 Basel, Switzerland.

Developmental Biology 397 (2015) 162–174

and an md neuron (Brewster and Bodmer, 1995; Orgogozo et al.,2001; Lai and Orgogozo, 2004). Of these lineages the one thatgives rise to the larval m-es organ and md neuron is best char-acterized (Orgogozo et al., 2001). Specifically, the SOP cell dividesto give rise to two secondary precursor cells, pIIa and pIIb. The pIIacell is the precursor of the socket and shaft cell, while the pIIb celldivides to generate an md neuron and the pIIIb cell, which furtherdivides to generate the sheath cell and a bipolar neuron. For p-esorgans, it has been proposed that the pIIIb cell goes throughadditional divisions to produce multiple neurons (Bodmer et al.,1989; Brewster and Bodmer, 1995).

The Pox neuro (Poxn) gene, a member of the Drosophila Pax genefamily, encodes a transcription factor with a DNA-binding paireddomain (Bopp et al., 1989; Noll, 1993). Poxn plays essential roles inmany aspects during the development of the PNS in both larvae andadults (Dambly-Chaudière et al., 1992; Awasaki and Kimura, 1997,2001; Boll and Noll, 2002). It is expressed in all larval p-es organs aswell as in a subgroup of m-es organs in a temporally and spatiallycontrolled fashion (Dambly-Chaudière et al., 1992; Awasaki andKimura, 2001; Boll and Noll, 2002). In Poxn null mutants, larval p-esorgans are transformed into m-es organs (Dambly-Chaudière et al.,1992; Awasaki and Kimura, 2001). These transformed mutant p-esorgans form papillae or hairs that are frequently shifted in position,whereby the shafts of the hairs are lost after the first larval instar.Thus, all transformed p-es organs appear as papilla-like m-es organsduring the second and third larval instars (Awasaki and Kimura, 2001;W.B., unpublished data). In addition, m-es hairs are transformed inPoxn null mutants to papilla-like organs that have no shafts, aphenotype that again appears only in second and third, but not infirst, instar larvae (Awasaki and Kimura, 2001; W.B., unpublisheddata). In adult Poxn null mutants, chemosensory sensilla are trans-formed into mechanosensory-like sensilla (Awasaki and Kimura, 1997;Boll and Noll, 2002). Null mutants of Poxn, PoxnΔM22, generated in ourlab, are viable, but males are sterile due to the absence of several Poxnfunctions each of which is regulated by a separate enhancer and isnecessary for male fertility (Boll and Noll, 2002; Krstic et al., 2009).

Although the importance of Poxn during the development of thep-es organs had been observed long ago, the function(s) of Poxnduring this process remained elusive. In this study, we describe novelcell lineages for all larval p-es organs, i.e., the thoracic kölbchen andtheir abdominal homologs, the lateral p6 and dorsal h3. In all wild-type larval p-es lineages, the SOP cells divide to generate twosecondary precursors, the pIIa and pIIb cells. Notably, in theselineages the pIIa cell is the precursor of all support cells, whereasthe pIIb cell gives rise to all neurons. This is in striking contrast to thelineage of m-es organs where the pIIb precursor generates bothepidermal – the sheath support cell – and neuronal cells. In addition,we find that one cell undergoes apoptosis in the wild-type p6lineage. On the other hand, Poxn null mutants show altered lineages,where the p-es division pattern is transformed into that of an m-eslineage. Moreover, apoptosis is no longer observed in any trans-formed p-es lineage of Poxn mutants. Together our results suggestthat Poxn plays a decisive role in the specification of the cell lineagesduring the development of p-es organs in Drosophila larvae.

Materials and methods

Fly stocks

The following fly strains were used:

y w,y w; E7-2-36 (Bier et al., 1989),w; UAS-p35 (kindly provided by Bruce Hay),y w; PoxnΔM22/CyO, yþ ,

w; PoxnΔM22/CyO; Poxn-CD8::GFP3-3,w; Poxn-Gal4ups1f/TM3,y w; UAS-CD8::GFP/CyO,y w; UAS-CD8::GFP; Poxn-Gal4ups1f/TM3,y w hs-flp; UAS4CD2 yþ4CD8::GFP/CyO; TM2/TM6B,y w hs-flp; UAS4CD2 yþ4CD8::GFP/CyO; Poxn-Gal4ups1f/TM6B,y w hs-flp; PoxnΔM22 UAS4CD2 yþ4CD8::GFP/CyO; Poxn-Gal4ups1f/TM6B,y w hs-flp; E7-2-36 UAS4CD2 yþ4CD8::GFP/CyO; Poxn-Gal4ups1f/TM6B,y w hs-flp; PoxnΔM22 E7-2-36/CyO; Poxn-Gal4ups1f/TM6B,y w hs-flp; PoxnΔM22 UAS-p35/CyO; Poxn-Gal4ups1f/TM6B.

Generation of flip-out clones in the embryonic PNS

Flies were crossed to generate a stock that combined threetransgenes (Fig. 2A): (i) hsp70-flp, the yeast recombinase gene flpunder control of the heat-inducible hsp70 promoter (Struhl andBasler, 1993); (ii) Poxn-Gal4ups1f (Poxn transgene EvK (Boll and Noll,2002), the coding region of which has been replaced by that of Gal4)that expresses Gal4 in all SOPs of larval p-es organs and their dau-ghter cells; and (iii) the reporter transgene UAS4CD2 yþ4CD8::GFP(derived from Actin5C4CD2 yþ4PKA* (Jiang and Struhl, 1995) andkindly provided by Gary Struhl). The stock was kept at 181C tominimize flp-induced recombination in the germline.

To collect embryos, flies were allowed to lay eggs on apple juice-containing plates, supplemented with a small amount of yeast, for2 h at 181C. Plates were kept at 181C for 4, 6, 8, 10, and 12 h, andembryos were collected in a fine custom-built nylon sieve. The age ofthe collected embryos thus corresponded to 2–3, 3–4, 4–5, 5–6, and6–7 h of development at 251C. Embryos were rinsed several timeswith cold tap water before they were heat-shocked by placing thenylon sieve into a pre-heated waterbath and keeping the embryossubmerged for 30 min. After heat shock under five different condi-tions (30 min or 15 min at 341C; 25 min or 30 min at 321C; 30 min at301C), the average numbers of GFP-labeled clones per embryo weredetermined (data not shown). A 30-min heat shock at 321C producedon average about one thoracic clone per embryo, thus labeling onlyone of the twelve kölbchen lineages, which was taken as anindication that only a single recombination event had taken placein this lineage. Therefore, these heat-shock conditions were usedthroughout our studies. After heat shock, embryos were rinsedbriefly with cold tap water and placed, still on the nylon sieve, intoa humidified chamber. Embryos were allowed to develop at 181Cuntil they reached what corresponds to 14–15 h of development at251C (mid stage 16) before they were fixed and immunostained.

Antibodies and immunohistochemistry

The following primary antibodies were used: rabbit anti-Poxn(1:50; Bopp et al., 1989); rabbit anti-D-Pax2 (1:100; Fu and Noll,1997); chicken anti-GFP (1:500; Abcam); mouse anti-CD2 (1:100;Serotec); rabbit anti-Su(H) (1:100; Santa Cruz Biotechnology); mouseanti-Prospero (1:10; batch MR1A, Developmental Studies HybridomaBank, University of Iowa (DSHB)); mouse anti-22C10 (1:100; DSHB);rabbit anti-β-galactosidase (1:1000; Cappel); rabbit anti-BarH1(1:100; Higashijima et al., 1992); rabbit anti-cleaved Casp-3(Asp175) (1:100; Cell Signaling Technology); and rat anti-Elav(1:200; DSHB). The secondary antibody used in Fig. 1 was biotiny-lated goat anti-rabbit IgG (VECTASTAIN ABC Kit, Vector LaboratoriesPK-4001); other secondary antibodies used in this study were: Alexa488-, Alexa 594-, and Alexa 647-conjugated anti-chicken, -rabbit,-mouse, or -rat IgG (1:500; all from Molecular Probes).

Embryos were fixed and stained as described previously(Gutjahr et al., 1993; Boll and Noll, 2002).

Y. Jiang et al. / Developmental Biology 397 (2015) 162–174 163

Quantification of cell number in kölbchen of Poxn null mutants

In PoxnΔM22 embryos, two types of kölbchen lineages wereobserved, producing five (Fig. 5A) or six cells (Fig. 5B). To testwhether this variation in cell number is an artifact of Gal4 thatinduces an additional mitosis in some of the lineages and to quantifyboth lineages, we analyzed the number of GFP-positive cells inPoxnΔM22; Poxn-CD8::GFP3-3 embryos in which expression of CD8::GFP (Lee and Luo, 1999) is controlled by the same Poxn promoter andenhancer as Poxn-Gal4ups1f (see above). In these embryos, again twotypes of kölbchen lineages were found of which 93/144 consisted offive and 51/144 of six cells, thus excluding the possibility that thevariation in cell number was the result of a Gal4 artifact.

Microscopy and image processing

Poxn protein was visualized in wild-type embryos by immuno-histochemistry and recorded with a Zeiss Axiophot microscope

equipped with a Hamamatsu CCD camera. Pictures of immuno-fluorescent proteins in clonal analyses were taken with a Leica TCSSP1 confocal microscope and a Zeiss LSM 710 confocal microscope.All immunofluorescent images are maximal projections of confocalZ-stacks that were processed with LCS Lite (Leica), Zen software(Zeiss), ImageJ, and Photoshop (Adobe Systems).

Results

Embryonic Poxn expression versus the Poxn mutant phenotype oflarval p-es organs

The expression of Poxn during embryonic development is highlydynamic (Dambly-Chaudière et al., 1992), as illustrated in Fig. 1 ingreater detail. Initially, a single dorsal cell of each thoracic andabdominal hemisegment expresses Poxn protein at late stage 10(Fig. 1A), followed by a second ventral cell during stage 11 (Fig. 1B).

Fig. 1. Expression of Poxn during embryogenesis. Poxn protein is visualized by staining with anti-Poxn antiserum in lateral views of whole-mount embryos at late stage 10(A), stage 11 (B), early (C) and late stage 12 (D), stage 13 (E), stage 14 (F), stage 15 (G), and stage 16 (H). Arrowheads in G and H indicate the Poxn-expressing bipolar neuronsinnervating the kölbchen. Embryos are oriented with their anterior to the left and dorsal side up. Abbreviations: T, thoracic segments; A, abdominal segments; dk, dorsalkölbchen; lk, lateral kölbchen; vk, ventral kölbchen; h3, hair 3; p6, papilla 6; br, brain; CNS, central nervous system; epid, epidermis.

Y. Jiang et al. / Developmental Biology 397 (2015) 162–174164

These cells are the SOPs of the larval p-es organs (Dambly-Chaudièreet al., 1992), which divide (Fig. 1B and C) and give rise to two clustersof Poxn-expressing cells per hemisegment by the end of germbandretraction (Fig. 1D). During germband retraction (stage 12), the dorsalclusters of Poxn-expressing cells in T2 and T3 are located ventrally tothose of the remaining segments (arrows in Fig. 1C and D). At theonset of stage 13, expression of Poxn begins to fade in the developingp-es organs (Fig. 1E). In the thoracic segments, however, one cell ofeach cluster continues to express Poxn very strongly at later stagesuntil the end of embryogenesis (Fig. 1F–H; arrowheads in panels Gand H). These single Poxn-expressing cells are bipolar neuronsinnervating the kölbchen. As evident from Fig. 1, Poxn is alsoexpressed in several cells of the gnathal and terminal abdominalsegments, in the ventral nerve cord and developing larval brain, in a

few cells that may belong to the olfactory and gustatory sensoryorgans of the larval head (not visible in focal planes of Fig. 1), in manyepidermal cells, and, after stage 14, in single cells associated withlarval mono-innervated hairs.

In larvae that are homozygous for the Poxn null allele, PoxnΔM22

(Boll and Noll, 2002), all p-es organs as well as a subset of m-esorgans exhibit a mutant phenotype. Of the p-es organs, thekölbchen and hair 3 are transformed into papilla-like sensoryorgans, and the transformed ventral kölbchen, papilla 6, and hair3 are mislocalized (data not shown). These defects are in agreementwith previous analyses of larvae that are homozygous for a largedeficiency, Df(2R)WMG, uncovering Poxn (Dambly-Chaudière et al.,1992), or a strong hypomorphic Poxn allele, Poxn70 (Awasaki andKimura, 2001).

Fig. 2. Strategy of marking clones and its application to the kölbchen lineage. (A) Labeling of flip-out clones in p-es organs in a time-dependent manner. Expression of CD2 indeveloping larval p-es organs is regulated by Gal4 under control of the Poxn enhancer and promoter in the Poxn-Gal4ups1f transgene (in the text abbreviated as Poxn-Gal4). Amoderate heat shock (hs) activates the hsp70 promoter that controls the yeast flp gene encoding FLP recombinase, which excises CD2 by recombination of flanking FRT sitesduring cell division and thereby brings CD8::GFP (green) under the control of Gal4. Note that cells of the p-es lineages in which no recombination was induced express CD2(red). ((B)–(P)) Labeled clones in dorsal ((I), (M) and (N)) or lateral (all other panels) kölbchen of stage 16 embryos are shown, which were induced at 2–3 ((B)–(F); except (E),where no clones were induced but all cells of a kölbchen were marked by CD2), 3–4 ((G) and (H)), 3–5 (I), 4–5 ((J) and (K)), and 5–6 h AEL ((L)–(P)) and immunostained withantibodies directed against the proteins indicated in the corresponding colors below the panels. Overlaps of D-Pax2 or Elav (blue) with Pros or β-gal (red) are purple ((D) and(I)), while those of GFP (green) with 22C10 (red) are yellow (K). Genotypes of embryos are y w hs-flp (homozygous or over Y); UAS4CD2 yþ4CD8::GFP (homozygous or overCyO); Poxn-Gal4ups1f (homozygous or over TM6B) and include (in (F), (I) and (P)) the enhancer trap marker E7-2-36 on the second chromosome. For a detailed description ofclones, see text. Here and in subsequent figures as well as in supplementary figures, panels include explanatory sketches that illustrate individual cells of the clones (derivedfrom 3D-analyses of the confocal pictures) and scale bars of 5 mm. Please note here and in subsequent figures that cells that are located outside of the flip-out clone and arealso labeled by the antibodies belong to either adjacent m-es organs or the same p-es organ, as e.g., in Fig. S2G.

Y. Jiang et al. / Developmental Biology 397 (2015) 162–174 165

Experimental strategy and cell lineage of kölbchen in wild-typeembryos

To address the role of Poxn during the development of larval p-esorgans, we first analyzed the cell lineage of kölbchen in wild-typeembryos. To this end, we designed an assay in embryos thatcombines temporally controlled FLP/FRT recombination (Golic andLindquist, 1989) with spatially controlled Gal4/UAS reporter expres-sion (Brand and Perrimon, 1993) to label flip-out clones (Struhl andBasler, 1993) in the lineage of larval p-es organs. These embryosexpress CD2 under the indirect control of a Poxn enhancer in all cellsof the larval p-es lineages until activation of the FLP recombinase by amoderate heat shock deletes CD2 in flip-out clones (Fig. 2A; Jiang andStruhl, 1995). Consequently, these clones express the fusion proteinCD8::GFP under the control of the Poxn enhancer (Fig. 2A). Twofeatures of this system were essential for proper analysis of larvalp-es organ lineages: (i) all cells of the lineages express Poxn, andhence the Poxn-Gal4 transgene, until stage 13 when cell division hasceased (Fig. 1); and (ii) the stability of the Gal4, CD2, and CD8::GFPproteins results in their perdurance, which permits clonal analysis atstage 16, when these cells have adopted their final fates and expresscell-type specific proteins. The combinatorial analysis of such cell-type specific markers and the patterns of CD8::GFP- and CD2-expressing cells in a large number of individual clones, induced atdifferent times after egg laying (AEL), enabled us to distinguish allcells produced after the last division of the lineage, to quantify eachtype of clone, and to derive the pattern of cell divisions.

When embryos were subjected to a moderate heat shock (30 minat 321C) during early embryogenesis (2–3 h AEL; although embryosdeveloped at 181C, except during heat shock, the indicated timeintervals throughout the paper correspond to development at 251C),flip-out clones were mostly produced before SOP selection (Table S1in Supporting information). Therefore, all cells that expressed CD8::GFP were derived from an SOP inwhich a flip-out event had occurred.Since such clones consisted of seven cells in dorsal or lateral kölbchen(dk or lk) (Fig. 2B) and in the ventral kölbchen (vk) (Fig. S1A inSupporting information), it follows that the kölbchen lineage givesrise to seven cells. Further detailed analysis was restricted to clones ofthe dk/lk lineages because their cells were better separated from eachother and hence much easier to distinguish by position (Fig. 2B) thanclones of the vk (Fig. S1 in Supporting information; see below).

The seven cells could be identified by the use of markers that areexpressed in a specific subset of larval es organ cells. The socket cellwas marked by the transcription factor Suppressor of Hairless [Su(H)] (Gho et al., 1996) and is always located most dorsally (Fig. 2C).The shaft cell and sheath cell expressed D-Pax2 (Fig. 2D), a paired-domain transcription factor homologous to Poxn (Fu and Noll, 1997;Hill et al., 2010). At the same time, the more ventrally locatedsheath cell also expressed the homeodomain transcription factorProspero (Pros) (Vaessin et al., 1991) (Fig. 2D). Finally, expression ofthe pan-neuronal nuclear protein Elav (Robinow and White, 1991)identified the remaining four cells as neurons (Fig. 2E), one ofwhich, consistently located at the most ventral position, was iden-tified as the md neuron (Fig. 2F) by its β-gal expression of theenhancer trap marker E7-2-36 (Bier et al., 1989; Brewster andBodmer, 1995). This result showed that the kölbchen lineagegenerates three bipolar neurons and one md neuron.

When embryos were heat-shocked at a slightly later stage duringembryogenesis (3–4 h AEL), the flip-out clones either labeled allseven cells or only a subset of these cells, the latter of which resultedfrom a FLP-induced recombination event after division of the SOP cell(Table S1 in Supporting information). When heat shock was appliedeven later (4–5 h AEL), nearly all flip-out clones labeled only a subsetof cells of the kölbchen lineage, among which mainly two types ofclones were observed (Table S1 in Supporting information). The firstconsisted of three cells, one of which was always located most

dorsally, while the remaining two expressed D-Pax2, which sug-gested that these were the three support cells (Fig. 2G). The secondtype consisted of four cells that did not express D-Pax2 and werelocalized ventrally of the CD2-expressing cells in the lineage, whichindicated that they were all neurons (Fig. 2H). This was confirmed bythe expression of Elav in these clones, one cell of which was an mdneuron that was consistently located most ventrally (Fig. 2I). Inaddition, observations with both types of clones were confirmed byimmunostaining for the axonal and dendritic marker 22C10(Hummel et al., 2000), which does not stain the first type (Fig. 2J)but does stain all cells of the second type (Fig. 2K). These resultssuggest that the socket, shaft, and sheath cells are derived from onesecondary precursor, pIIa, whereas all neurons are derived from theother, pIIb (Fig. 7A).

To determine the cell division patterns of the secondaryprecursors, pIIa and pIIb, clones were induced at 5–6 h AEL. Fouradditional types of flip-out clones were observed (Table S1 inSupporting information), consisting of (i) a single CD8::GFP-positive cell, which is the socket cell, evident from its most dorsalposition (Fig. 2L); (ii) the shaft and sheath cells, based on theirexpression of D-Pax2 (Fig. 2M); (iii) two bipolar neurons, accord-ing to their positions (neither most dorsal nor most ventral in thelineage) and the lack of D-Pax2 expression (Fig. 2N); and (iv) abipolar neuron and the md neuron, since the latter was alwayslocated most ventrally (Fig. 2O). As shown in Fig. 1, Poxn continuesto be expressed in a single cell of the kölbchen lineage throughoutembryogenesis (Fig. 1F–H). This Poxn-expressing cell is a bipolarneuron, as it formed a clone with the md neuron (Fig. 2P). Theremaining two bipolar neurons could not be distinguished fromeach other by the expression of BarH1, unlike in the lineage ofpapilla p6 (see below), because although BarH1 is stronglyexpressed in the outer support cells, the socket and shaft cells,and a bit less in the inner support cell, the sheath cell, it is notexpressed in any of the kölbchen neurons (Y.J., unpublished data).

Because individual cells could not be resolved in most cases, itwas impossible to analyze the vk lineage in the same detail as thatof dk/lk. However, in cases where the positions of individual cellscould be distinguished (Fig. S1 in Supporting information), the vkand dk/lk lineages are identical. Thus, vk also consisted of sevencells, four of which were labeled with Elav, and hence are neurons,while the remaining three were support cells (Fig. S1A in Support-ing information). Like in the dk/lk lineage, the vk support cellswere derived from one secondary precursor, pIIa, and two of themexpressed D-Pax2 (Fig. S1B in Supporting information), whereasthe four vk neurons were derived from the other secondaryprecursor (Fig. S1C in Supporting information), pIIb. In addition,one of the bipolar neurons of the vk continues to express Poxn(Fig. S1D in Supporting information and Fig. 1F–H), while one is anmd neuron (Fig. S1D in Supporting information).

Surprisingly, the lineage of kölbchen derived from this analysis(Fig. 7A) differs drastically from that reported for one of its twoabdominal homologs, papilla 6 (p6) (Brewster and Bodmer, 1995;Fig. 7D), which was thought to be similar to the lineage of m-esorgans (Orgogozo et al., 2001; Fig. 7E). Therefore, we also analyzedthe p6 lineage by the same method we used to determine thekölbchen lineage.

Cell lineage of papilla 6 in wild-type embryos

The p6 lineage had been proposed to generate six cells, threesupport cells, two bipolar neurons, and one md neuron (Brewsterand Bodmer, 1995; Fig. 7D). Yet, most flip-out clones induced at 2–3or 3–4 h AEL consisted of nine cells (Fig. 3A and Table S2 inSupporting information), three of which were located at a ventraland posterior position (marked by brackets in Fig. 3A). These ninecells formed a single clone that comprises the entire lineage

Y. Jiang et al. / Developmental Biology 397 (2015) 162–174166

because (i) the vast majority of marked clones consisted of ninecells when they were labeled by heat shock long before theselection of SOPs (2–3 h AEL in Table S2 in Supporting information),(ii) heat-shock conditions had been optimized to generate only onerecombination event per thorax, and (iii) no CD2-positive cells wereobserved in these clones (data not shown). The three ventral cells ofthe clone have been mentioned previously and were suggested toform an epidermal gland (Orgogozo and Schweisguth, 2004). Theexpression of GFP in these cells is not the result of ectopicexpression of Poxn-Gal4, as Poxn is also expressed in all three cells,which are distinct from the p6 cells at stage 13 (data not shown).While none of these three presumptive epidermal gland cellsexpressed the neuronal markers 22C10 or Elav, two did expressD-Pax2 at late embryogenesis (data not shown). When embryoswere heat-shocked later (3–4 h AEL in Table S2 in Supportinginformation), about half of the flip-out clones consisted in equalparts of all p6 (Fig. 3B) or all presumptive epidermal gland cells(Fig. 3C). These results suggest that the SOP of p6, pI, and theprecursor of the presumptive epidermal gland, pI’, are derived fromthe same progenitor cell p0 that expresses Poxn (Fig. 3D). In thefollowing, we restricted our analysis to the p6 lineage.

In agreement with Brewster and Bodmer (1995), we observedthat the p6 lineage gives rise to six cells. Similar to kölbchen, thep6 socket cell was always located most dorsally, the shaft andsheath cells of p6 expressed D-Pax2, while the three p6 neurons

were located more ventrally and expressed the neuronal marker22C10 (Fig. 3E). However, when heat shock was applied at 5–6 h AELto mark clones that include only part of the p6 lineage (Table S2 inSupporting information), most of the clones consisted of either allsupport cells (Fig. 3F) or all neurons (Fig. 3G and H), again as we haveobserved for kölbchen. One p6 neuron expressed E7-2-36 and hencewas an md neuron (Fig. 3G and H), while only one of the two bipolarneurons expressed BarH1 (Fig. 3H) and thus can be distinguishedfrom the other.

To examine the division patterns of the pIIa and pIIb cells of p6,clones were induced at 6–7 h AEL, which generated mainly fouradditional types of clones (Table S2 in Supporting information). Thefirst included only the socket cell, as it was located most dorsally andexpressed neither D-Pax2 nor 22C10 (Fig. 3I); the second consisted ofthe D-Pax2-expressing shaft and sheath cells (Fig. 3J); the third of thetwo bipolar neurons that express 22C10 but not the md neuronalmarker (Fig. 3K); while the fourth included only the md neuron(Fig. 3L). Thus, the p6 lineage derived from these results is identical tothe kölbchen lineage with the exception that the latter includes anadditional bipolar neuron (Fig. 7A).

Apoptosis in the p6 lineage

This difference between the kölbchen and p6 lineages couldarise either from an additional division in the kölbchen lineage or

Fig. 3. Clones of the papilla 6 lineage. ((A)–(C) and (E)–(L)) Marked clones of stage 16 embryos are shown that were generated at 3–4 ((A)–(C)), 5–6 ((E)–(H)), and 6–7 h AEL((I)–(L)) and immunostained with antibodies directed against the proteins indicated in the corresponding colors below the panels. Overlaps of GFP (green) with 22C10 (red)are yellow ((E), (G), (K) and (L)). Genotypes of embryos are as described in the legend to Fig. 2. For a detailed description of clones, see text. (D) Lineage of progenitor cell p0that, upon division, generates the SOP, pI, of the p6 lineage and the precursor, pI’, of the presumptive epidermal gland (Orgogozo and Schweisguth, 2004).

Y. Jiang et al. / Developmental Biology 397 (2015) 162–174 167

from apoptosis of one of the bipolar neurons in the p6 lineage. Ifthere is apoptosis in the p6 lineage, it is expected to occur duringcell differentiation between the ends of stages 12 and 15. There-fore, stage 12–15 embryos whose p-es organs were labeled byUAS-CD8::GFP under the control of Poxn-Gal4 were tested forapoptosis by antibodies that recognized activated Caspase-3 (Casp-3;Yu et al., 2002). Indeed, in about 70% of the cases (43/62), a GFP-labeled cell adjacent to the p6 cells also stained for activated Casp-3 instage 13 embryos (asterisk in Fig. 4A). This cell was supernumerary,since all other cells of the p6 and the epidermal gland lineages(bracket in Fig. 4A) were present. Moreover, when apoptosis wasinhibited by the apoptosis inhibitor P35 of baculovirus (Hay et al.,1994), the combined p6 and epidermal gland lineages produced tenrather than nine cells (Fig. 4B).

To determine whether the apoptotic cell belongs to the p6 orepidermal gland lineage (Fig. 3D), clones were induced to mark theentire p6 lineage or a sub-lineage of it in embryos that express P35in the lineage. In clones that labeled only the p6 lineage, asupernumerary cell was present, which in many cases (34/55)also expressed the neuronal marker Elav (Fig. 4E), but not the mdneuronal marker E7-2-36 (Fig. 4C). By contrast, no extra cell wasobserved in clones that consist only of the epidermal gland cells(data not shown). Consistent with these observations, the super-numerary cell was found in clones that label only the neurons(asterisk in Fig. 4D). Finally, the smallest marked clones thatincluded the supernumerary cell also labeled the md neuron(Fig. 4E), which demonstrates that these cells are derived fromthe same precursor. It follows that the p6 and kölbchen lineagesare identical, except that the bipolar neuron, which continues to

express Poxn in the kölbchen lineage (Fig. 7A), undergoes apop-tosis in the p6 lineage (Fig. 7B).

Cell lineage of hair 3 in wild-type embryos

Finally, we examined the lineage of the dorsal abdominal larvalp-es organ, h3. All flip-out clones induced in its primary precursor,pI or SOP, consisted of six cells, three of which were neuronsexpressing Elav, while two support cells expressed D-Pax2 (Fig. S2Ain Supporting information), as was observed for p6 (Fig. 3E).However, in contrast to p6 whose three support cells are all locateddorsally to the three neurons (Fig. 3E and F), the h3 shaft and sheathcells that express D-Pax2 are located ventrally to the two bipolarneurons (Fig. S2A in Supporting information). Clones larger than sixcells, like those of the p6 and epidermal gland lineage (Fig. 3A andTable S2 in Supporting information), were never observed for h3,when induced as early as 3–4 h AEL. When flip-out was induced ata later stage in the h3 secondary precursors, pIIa or pIIb, two typesof three-cell clones were observed, which were either all supportcells (Fig. S2B in Supporting information) or all neurons (Fig. S2Cand D in Supporting information). In the first case, only the mostdorsal cell did not express D-Pax2 and was likely the socket cell,while the other cells were shaft and sheath cells. Thus, like in the p6lineage, all three support cells of h3 are derived from the secondaryprecursor, pIIa, while all neurons are descendants of the secondaryprecursor, pIIb. Likewise, the most ventral neuron in the h3 lineagewas always the md neuron (Fig. S2D in Supporting information).Flip-out clones generated even later, in tertiary precursors, includedtwo types of two-cell clones: the D-Pax2-expressing shaft and

Fig. 4. One cell of the p6 lineage undergoes apoptosis. (A) An apoptotic cell marked by activated Casp-3 (asterisk) is shown among ten GFP-labeled p6 and epidermal gland(bracket) cells in a y w; UAS-CD8::GFP; Poxn-Gal4ups1f (homozygous or over TM3) embryo at stage 13. The interpretation in the sketch of the staining pattern was derived froman analysis of suitable substacks of confocal laser scanning microscope sections. The sketch does not include the three epidermal gland cells. (B) Inhibition of apoptosis byP35 generates a supernumerary cell that is associated with p6 and epidermal gland (bracket) cells of a y w hs-flp/(w or Y); UAS4CD2 yþ4CD8::GFP/UAS-p35; Poxn-Gal4ups1f/þembryo at stage 16. The sketch does not include the three epidermal gland cells. ((C)–(E)) Clones of stage 16 y w hs-flp/(w or Y); E7-2-36 UAS4CD2 yþ4CD8::GFP/UAS-p35;Poxn-Gal4ups1f/þ embryos, expressing P35 in p-es organs, were marked at 5–7 h AEL and immunostained with antibodies directed against the proteins indicated in thecorresponding colors below the panels. Overlaps of Elav (red) with β-galactosidase of E7-2-36 (blue) are purple ((C)–(E)). Asterisks indicate the supernumerary cell. For adetailed description of clones, see text.

Y. Jiang et al. / Developmental Biology 397 (2015) 162–174168

sheath cells (Fig. S2E in Supporting information) or the two bipolarneurons (Fig. S2F in Supporting information) that are located fardorsally to the md neuron (Fig. S2D in Supporting information). Itfollows that the shaft and sheath cells are generated from thetertiary precursor pIIIa, while the two bipolar neurons, the moreventral of which weakly expresses BarH1 (Fig. S2G in Supportinginformation), are derived from the tertiary precursor pIIIb. Thus, thelineage of h3 (Fig. 7C) is similar to the lineage of p6 (Fig. 7B), butdifferent in two important aspects. Although division of pIIb alsoproduces the tertiary precursor pIIIb that, like pIIIb1 in the p6lineage, generates two bipolar neurons, pIIb does not give rise to asecond tertiary precursor, since the second daughter cell of pIIbdifferentiates into an md neuron. In addition, apoptosis was neverobserved in the h3 lineage when it was analyzed for activated Casp-3 at stages 12–15 (Fig. S3 in Supporting information).

No apoptosis in the kölbchen lineage

To test whether apoptosis occurs in the kölbchen lineage, weanalyzed GFP-labeled kölbchen cells for the presence of apoptosisin stage 12–15 embryos. None of these cells stained for activatedCasp-3 (Fig. S4A and B in Supporting information, and data notshown). In agreement with this result, we never observed morethan seven kölbchen cells when P35 was used to inhibit cell death

in their lineage (Fig. S4C in Supporting information). Thus, unlikein the p6 lineage, apoptosis plays no role in the kölbchen lineage ofwild-type embryos.

Alteration of the kölbchen lineage in Poxn null mutants

Since larval p-es organs are transformed into mechanosensory-like sensilla in the absence of Poxn (Dambly-Chaudière et al.,1992), we asked how this affected their lineages. In Poxn nullmutants, the ventral kölbchen localized more dorsally and its cellswere much better resolved from each other. Therefore, all kölb-chen were included in this lineage analysis. We initially inducedclones that labeled the entire lineage. Surprisingly, we observed indorsal/lateral as well as ventral kölbchen two types of lineages,both of which generated a reduced number of differentiating cells:Two thirds (93/144) of the mutant kölbchen produced only fivecells (Fig. 5A), whereas the rest (51/144), only six cells (Fig. 5B).The five cells consisted of a most dorsally located Su(H)-expressingsocket cell (Fig. 5C), a D-Pax2-positive shaft cell (Fig. 5D), a moreventrally located sheath cell that expressed both D-Pax2 and Pros(Fig. 5D), and two neurons marked by 22C10 (Fig. 5E). When thelineage produced six cells, these included the same support cells(data not shown) and an additional neuron (Fig. 5F). Like in wild-typeembryos, the most ventrally located neuron of both lineages was an

Fig. 5. Altered kölbchen lineage in Poxn null mutants. Clones of stage 16 Poxn mutant embryos were marked at 3–4 ((A)–(F)), 4–5 ((G)–(I)), and 5–6 h AEL ((J)–(L)) andimmunostained with antibodies directed against the proteins indicated in the corresponding colors below the panels. Overlaps of GFP (green) with 22C10 (red) are yellow((E) and (F)), those of D-Pax2 (blue) with Pros (red) are purple ((D) and (I)). Genotypes of embryos are y w hs-flp (homozygous or over Y); PoxnΔM22 UAS4CD2 yþ4CD8::GFP;Poxn-Gal4ups1f (homozygous or over TM6B) ((A)–(D) and (G)–(L)) or y w hs-flp; PoxnΔM22 UAS4CD2 yþ4CD8::GFP/PoxnΔM22 E7-2-36; Poxn-Gal4ups1f (homozygous or overTM6B) ((E) and (F)). For a detailed description of clones, see text.

Y. Jiang et al. / Developmental Biology 397 (2015) 162–174 169

md neuron, as it expressed the E7-2-36 marker (Fig. 5E and F). Thus,while all transformed kölbchen of PoxnΔM22 mutants had threesupport cells and an md neuron, they also showed a reduced numberof one or two bipolar neurons.

To analyze the pattern of cell divisions in the mutant kölbchenlineage, clones were labeled at 4–5 h AEL. Two types of clones thatdid not label the entire lineage were observed (Table S3 in Supportinginformation). One consisted of socket cell and shaft cell, based ontheir position within the lineage and the expression of D-Pax2(Fig. 5G), respectively. The other type of clone labeled the D-Pax2-expressing sheath cell and two (Fig. 5H) or three neurons (data notshown). Cell identities in these clones were corroborated by stainingfor Pros, which was absent in the socket and the D-Pax2-expressingshaft cell of the first type of clone (data not shown), but was presentin the D-Pax2-expressing sheath cell of the second type (Fig. 5I). Theremaining two cells of this second type of clonewere neurons, as theywere located more ventrally and expressed neither D-Pax2 nor Pros(Fig. 5I). The same types of clones were revealed in the mutantlineage that produced six cells (data not shown). These results showthat in the kölbchen lineage of PoxnΔM22 mutants the socket and shaftcell are daughters of the pIIa cell, while the pIIb cell generates thesheath cell and neurons (Fig. 7F).

The division pattern of the pIIb cell was further examined inclones induced at 5–6 h AEL. Three types of clones were informa-tive (Table S3 in Supporting information). One type, which ispresent in both five- and six-cell lineages of mutant kölbchen,consisted of the sheath cell and one bipolar neuron, which islocated dorsal to the md neuron (Fig. 5J); a second type consistedonly of the most ventrally located md neuron, when the lineageproduced five cells (Fig. 5K); and a third type consisted of a bipolarneuron and the md neuron, when the lineage produced six cells(Fig. 5L). Thus, Poxn mutant kölbchen appear in two forms whichresult from two similar lineages: (i) when the organ consists of sixcells, its lineage resembles that of an m-es organ (Fig. 7E), in whichthe precursor of the md neuron undergoes an additional divisionto produce a second bipolar neuron (Fig. 7F); (ii) when thetransformed organ includes only five cells (Fig. 7G), its lineage isidentical to that of an m-es organ (Fig. 7E).

To determine whether the reduction in cell number in theabsence of Poxn results from cell death, we examined the kölbchenlineage in homozygous PoxnΔM22 embryos for apoptotic cells. Thus,all p-es cells of stage 12–15 embryos were labeled with antibodiesagainst CD2 (Fig. 2A) and stained for activated Casp-3. As themutant kölbchen consisted of at least five cells, we focused onlineages consisting of at least six CD2-positive cells. However, weobserved neither co-expression of activated Casp-3 with CD2 at anystage nor any mutant kölbchen lineages consisting of more than sixcells (Fig. S5A and B in Supporting information, and data notshown). Consistent with this result, when UAS-p35 was expressedunder the control of Poxn-Gal4 in the kölbchen lineage of PoxnΔM22

embryos, we found the same types of lineages described above(Fig. 5A and B), consisting of five (18/26; Fig. S5C in Supportinginformation) or six cells (8/26; Fig. S5D in Supporting information)with a ratio (18/8) similar to that observed in embryos that did notexpress P35. These results suggested that the reduced number ofcells in the kölbchen lineage of Poxn null mutant embryos resultedfrom a reduced number of cell divisions rather than from apoptosis.

Transformation of abdominal p-es organ lineages to an m-es organlineage in the absence of Poxn protein

We next analyzed the p6 lineage of PoxnΔM22 mutants. Earlyinduced clones labeled the entire lineage and included the pre-sumptive epidermal gland cells (bracket in Fig. 6A). The mutant p6cells were located more dorsally and showed much more overlapwith each other (Fig. 6A) than in wild-type embryos (Fig. 3A). Thelineage generated five differentiating cells: a socket cell, twoD-Pax2-expressing cells—the shaft and sheath, and two Elav-expressing neurons (Fig. 6B). When the entire lineage was labeledwith CD2 and embryos were stained for Casp-3 during stages 13 to15 (similar to wild-type p6 in Fig. 4A), no apoptosis was observed inthe lineage. Therefore, the reduced number of five cells does notresult from apoptosis, but rather from a reduced number of celldivisions. To determine the division pattern of the mutant p6lineage, clones were induced after division of the SOP to label sub-lineages. Two types of clones were found: one consisted of two cells,

Fig. 6. Altered p6 lineage in Poxn null mutants. Clones of stage 16 Poxn mutant embryos were marked at 3–4 (A), 4–5 ((B)–(F)), and 5–6 or 6–7 h AEL ((G) and (H)) andimmunostained with antibodies directed against the proteins indicated in the corresponding colors below the panels. Overlaps of Elav or Pros (red) with D-Pax2 orβ-galactosidase of E7-2-36 (blue) are purple ((E), (F) and (H)). Genotypes of embryos are as described in the legend to Fig. 5. For a detailed description of clones, see text.

Y. Jiang et al. / Developmental Biology 397 (2015) 162–174170

the socket cell and the D-Pax2-expressing shaft cell, neither ofwhich expressed Elav (Fig. 6C); the other consisted of three cells, theD-Pax2-expressing sheath cell and two neurons (Fig. 6D), one ofwhich was identified as the md neuron through its expression of theE7-2-36 marker (Fig. 6E). The identity of the sheath cell in three-cellclones was again confirmed by its expression of both Pros and D-Pax2 (Fig. 6F). These results suggest that in the p6 lineage ofPoxnΔM22 mutants, the socket and shaft cell are produced by thepIIa cell, while the sheath cell and neurons are derived from the pIIbcell (Fig. 7G). Two types of clones were observed in the pIIb lineage:one consisted of the D-Pax2-expressing sheath cell and an Elav-expressing neuron (Fig. 6G); the other type consisted only of the mdneuron, which is marked by E7-2-36 (Fig. 6H). Since no two-cellclone expressed the md neuron-specific marker, it follows that the

sheath cell and the bipolar neuron are derived from pIIIb. Therefore,in Poxn mutants the p6 lineage (Fig. 7G) is identical to that of wild-type m-es organs (Fig. 7E).

Finally, the lineage of the dorsal abdominal p-es organ, h3, wasanalyzed in PoxnΔM22 mutants. Early induced flip-out clonesconsisted of only five cells, instead of the six observed in thewild-type h3 lineage (cf. Fig. S2H with A in Supporting informa-tion). Again staining for Casp-3 in the mutant h3 lineage markedby CD2 revealed no apoptosis, like in the mutant lineages of p6and kölbchen. Two of these five cells were neurons expressingElav, two were shaft and sheath cells that express D-Pax2, and oneexpressed neither of these markers and was the socket cell, since itwas located most dorsally (Fig. S2H in Supporting information).Like in the mutant p6 lineage, two types of two-cell mutant h3

Fig. 7. Lineages of larval p-es organs and their alterations in Poxn mutants. (A) Lineage of wild-type kölbchen. (B) Lineage of wild-type papilla p6. (C) Lineage of wild-typehair h3. (D) Lineage previously proposed for wild-type papilla p6 (Brewster and Bodmer, 1995). (E) Lineage of larval m-es organs (Orgogozo et al., 2001). ((F) and (G))Alternative kölbchen lineages ((F) and (G)) and h3 or p6 lineage (G) in Poxn null mutants. The neuron with a white center (F) corresponds to the neuron that would expressPoxn in the wild-type kölbchen lineage (A). Different colors of cells derived from the same lineage are to indicate that the cells could be distinguished from each other by theexpression of suitable markers: the socket, the most dorsal cell (except in h3) expresses Su(H) (assumed to be true also for h3), but not D-Pax2; shaft and sheath cells expressD-Pax2, but not Su(H), while the sheath cell also expresses Pros (assumed to be true also for h3); all neurons express Elav, while the most ventral cell, the md neuron, is theonly cell that expresses the enhancer trap marker E7-2-36. The socket, shaft, and sheath cells, and neurons are each represented in the same colors to indicate same celltypes in different lineages. The bipolar neuron, whose sibling is the md neuron and which is present only in the kölbchen lineage, is the only neuron that expresses Poxn.BarH1 is expressed in socket, shaft, and sheath cell but in none of the neurons of the kölbchen lineage. By contrast, BarH1 is expressed in one of the two bipolar neurons andin the presumptive sheath cell of the p6 and h3 lineages. For discussion, see text.

Y. Jiang et al. / Developmental Biology 397 (2015) 162–174 171

clones were observed. One type consisted of two support cellsbecause neither of them expressed Elav (Fig. S2I in Supportinginformation), while one of them expressed D-Pax2 (Fig. S2I inSupporting information) but not Pros (Fig. S2J in Supportinginformation). Hence, the D-Pax2-expressing cell was the shaft cell,and the other cell that expressed neither D-Pax2 nor Pros was thesocket cell. The other type of two-cell mutant h3 clone consisted ofa D-Pax2-expressing cell and an Elav-expressing cell, which thusare the sheath cell and a bipolar neuron, respectively (Fig. S2L inSupporting information), because none of the two-cell clones everexpressed the md neuron-specific marker E7-2-36. The only typeof three-cell mutant h3 clone observed consisted of two Elav-expressing neurons and one D-Pax2-expressing cell, which thuswas the sheath cell (Fig. S2K in Supporting information). The moreventral of the two neurons displayed multiple dendrites and thusis an md neuron (Fig. S2K in Supporting information). Thus in Poxnmutants, the lineage of h3, like that of p6, is transformed to that ofm-es organs (Fig. 7G).

Discussion

The lineages of the larval es organs may appear simple, as theyinclude only few cells. The simplest is the m-es organ lineage,which generates four cells, in addition to an md neuron (Orgogozoet al., 2001; Fig. 7E). In comparison, the larval p-es organ lineagesare slightly more complex, such as that of the abdominal papillap6, which has an additional bipolar neuron. Despite their simpli-city, however, essential features of the p-es lineages determinedhere (Fig. 7A–C) deviate from those proposed (Bodmer et al., 1989)or determined by an earlier analysis (Brewster and Bodmer, 1995;Fig. 7D). For example, in contrast to the m-es lineage (Orgogozoet al., 2001; Fig. 7E) and the previously reported p6 lineage(Fig. 7D), we have demonstrated that the three support cells ofall p-es lineages are in fact derived from one secondary precursor,while the other secondary precursor generates only neurons andthus is a purely neuronal precursor (Fig. 7A–C). Moreover, theprimary precursor cell generates the entire p-es organ in onlythree (Fig. 7A–C) rather than the four rounds of cell division thatwere thought to be necessary (Brewster and Bodmer, 1995;Fig. 7D). In addition, we have shown that, different from all otherp-es lineages, the p6 lineage involves apoptosis (Fig. 7B). Surpris-ingly, the p6 lineage also differs in one important aspect from thatof its dorsal analog, the abdominal p-es organ, hair h3, eventhough the same types and number of cells are produced in bothlineages (Fig. 7B and C). In the h3 lineage, the neuronal secondaryprecursor, pIIb, divides to generate an md neuron and theprecursor of two bipolar neurons (Fig. 7C). In the lineage of p6,however, pIIb produces two precursors: one that generates twobipolar neurons, and another that produces an md neuron and acell determined to undergo apoptosis (Fig. 7B). Thus, in the formerlineage the absence of one cell division is compensated by theabsence of apoptosis, whereas in the latter lineage apoptosisannihilates the additional cell division. We have further shownthat the lineage of the thoracic homologs of the abdominal p-esorgans, the kölbchen, is identical to that of the p6 lineage, exceptthat the neuronal secondary precursor in the kölbchen lineageproduces three, instead of only two, bipolar neurons becauseapoptosis does not occur (Fig. 7A). Finally, analysis of the p-esorgan lineages in the absence of Poxn has revealed the importanceof Poxn functions at various levels of these lineages.

Differences and similarities between cell lineages of larval p-es organs

In contrast to the cell lineages of m-es organs that have beenextensively studied in both larvae and adults (Bodmer et al., 1989;

Hartenstein and Posakony, 1989; Brewster and Bodmer, 1995; Ghoet al., 1999; Reddy and Rodrigues, 1999; Orgogozo et al., 2001;Fichelson and Gho, 2003), little was known about the develop-ment of p-es organs. In this study, we have described the celllineages for all larval p-es organs, the thoracic kölbchen as well astheir abdominal homologs, h3 and p6 (Fig. 7A–C). Interestingly,these lineages are quite distinct from any previously reportedlineages of sensory organs in Drosophila.

All p-es organ lineages share one part that is identical, namelythe lineage derived from one of the two secondary precursors,pIIa (Fig. 7A–C). This precursor generates the three support cells,the socket cell in a first division and the shaft and sheath cells ina subsequent division. On the other hand, the p-es lineagesderived from the secondary precursor pIIb are similar but showclear differences. In the dorsal abdominal p-es organ, h3, pIIbdivides to generate the md neuron and one tertiary precursor,pIIIb (Fig. 7C). However, in the thoracic kölbchen and the ventralabdominal homolog p6, pIIb produces two tertiary precursors,one of which, pIIIb1, divides to form two bipolar neurons, whilethe other, pIIIb2, gives rise to an md neuron and a third bipolarneuron in the kölbchen lineage (Fig. 7A) or to an md neuron and acell that is eliminated by apoptosis in the p6 lineage (Fig. 7B).Therefore, the kölbchen has three bipolar neurons, whereas p6has only two.

There are additional important differences between thelineages of p6 and kölbchen. First, the SOP of p6, pI, is not derivedfrom a proneural cluster but from a precursor, p0 (Fig. 3D), thatalso generates a presumptive epidermal gland (Orgogozo andSchweisguth, 2004). This common precursor, p0, is selected froma proneural cluster by lateral inhibition and begins to expressPoxn. Second, the sibling bipolar neuron of the md neuron in thekölbchen lineage is the only cell in all p-es organ lineages thatcontinues to express Poxn throughout embryogenesis. Third, oneof the bipolar neurons of p6 expresses BarH1, a gene that is notactive in the bipolar neurons of kölbchen (data not shown).

It is unclear how an earlier analysis of the p6 lineage, in whichthe clones were labeled by flip-out in the act4Draf4nuclacZtransgene (Struhl and Basler, 1993) and neurons were identified bystaining for 22C10 (Brewster and Bodmer, 1995), led to a modelthat is similar to the m-es lineage (Brewster and Bodmer, 1995;Fig. 7D). The analysis that led to this model differs from ours in atleast one important aspect. In this earlier study, the transgene isexpressed ubiquitously after flip-out. Thus one might argue thatmarked clones of epidermal cells adjacent to the p6 cells couldhave interfered with the analysis of the p6 clones. However, sincethe frequency of clones observed in the PNS was below three perembryo (Brewster and Bodmer, 1995), it seems improbable thatsuch an effect could explain the 12 clones observed that wereproposed to consist of the sheath cell and all neurons, or the5 clones that were thought to consist of only the socket and shaftcells (Brewster and Bodmer, 1995). Neither could such interferenceexplain that no clones were observed that consisted only of thethree support cells (Brewster and Bodmer, 1995), which wouldhave been expected to occur at a high frequency similar to that ofclones including all neurons. Since these results were only docu-mented by a table, it appears probable that this study was guidedby adhering to a lineage previously proposed for larval p-es organs(Bodmer et al., 1989).

Multiple functions of Poxn in the lineages of p-es organs

In homozygous PoxnΔM22 larvae, p-es organs are transformedinto papilla-like mono-innervated organs or hairs that, like allm-es hairs, lose their shaft after the first instar. Our analysis of allp-es organ lineages, in the wild type and Poxnmutants, shows thatnot only the phenotypes but also the lineages of all p-es organs are

Y. Jiang et al. / Developmental Biology 397 (2015) 162–174172

transformed (Fig. 7G) into that of an m-es organ (Fig. 7E). This is ifwe disregard a minority of a third of all kölbchen lineages thatslightly deviates from the m-es lineage by forming a sibling of themd neuron, which is a second bipolar neuron (Fig. 7F). Thus, wemay consider the formation of an m-es organ as the defaultdevelopmental pathway in the absence of Poxn. In this case, onewould expect that ectopic expression of Poxn in the m-es organlineage is able to transform m-es into p-es organs. Although thisquestion has not been addressed experimentally, a similar experi-ment is consistent with such a conclusion: high levels of ubiqui-tous Poxn expression for 15 min during stages 9 to 11 is able toproduce ectopic kölbchen in the thorax (Dambly-Chaudière et al.,1992). Based on the sum of our results, our interpretation of thisubiquitous Poxn expression phenotype is more plausible than thealternative explanations that (i) ectopic Poxn recruits additionalSOPs and thus gives rise to ectopic kölbchen, (ii) ectopic Poxn cantransform internal sensory, i.e., chordotonal, organs into super-numerary kölbchen, or (iii) high levels of Poxn in the kölbchenlineage are able to duplicate kölbchens. Therefore, Poxn is not onlynecessary, as shown here, but may also be sufficient to discrimi-nate between m-es and p-es development.

These considerations also argue that the differences betweenthe lineages of kölbchen (Fig. 7A) and their abdominal homologs,p6 (Fig. 7B) and h3 (Fig. 7C), are determined by factors differentfrom Poxn, such as those encoded by the bithorax complex which,if absent, transform the abdominal h3 and p6 at least partially intokölbchen (Dambly-Chaudière and Ghysen, 1987). Similarly, it isprobable that the difference between the lineages of p6 (Fig. 7B)and h3 (Fig. 7C) result from factors expressed differentially alongthe dorsoventral axis of the embryo.

Keeping this in mind, it should be noted that the pIIa part of thep-es lineages is always changed in the same manner to the m-estype in Poxn mutants (Fig. 7E–G). In the wild-type p-es lineage,pIIa produces all support cells, including the sheath cell, whereasin the Poxn mutant lineages, pIIa generates only the socket andshaft cells, like in the m-es lineage. We conclude that Poxnincreases the developmental potential of pIIa and pIIIa, andprevents differentiation by inducing mitosis in pIIIa. Analogousconsiderations further suggest that Poxn reduces the develop-mental potential of pIIb and its offspring, which is restricted toneural fates in the presence of Poxn. However, in the absence ofPoxn, the developmental potential of pIIb and its offspring isexpanded and similar or identical to those of pIIb and pIIIb of them-es lineage, which are able to form a sheath cell in addition toneurons. In pIIIb2, which is not produced in h3 (Fig. 7C), Poxninhibits differentiation and promotes mitosis, thereby generatingan md neuron and a bipolar neuron in kölbchen (Fig. 7A) or an mdneuron and an apoptotic cell in p6 (Fig. 7B). In the absence of Poxn,however, pIIIb2 is always absent in the transformed lineages of p6and h3 (Fig. 7G) and is generated in only a third of the kölbchen(Fig. 7F). This shows that without Poxn pIIb can still produce pIIIb2in a fraction of the kölbchen lineage, which is then able to promotemitosis and form a bipolar neuron and an md neuron. It followsthat in this sublineage of the kölbchen Poxn acts with at leastanother factor to regulate its specification.

While the pIIIb2-generated bipolar neuron in the kölbchenlineage continues to express Poxn under wild-type conditions(Fig. 7A), the analogous cell in the p6 lineage does not, but insteadundergoes apoptosis. One might therefore hypothesize that Poxninhibits apoptosis of this neuron in the kölbchen lineage. However,the fact that this bipolar neuron does not undergo apoptosis whenpIIIb2 is formed in a third of all mutant kölbchen that lack Poxn(Fig. 7F), argues strongly against such a hypothesis. Rather, wepropose that the continued expression of Poxn in this bipolarneuron of the kölbchen lineage is necessary for its properspecification and differentiation. Interestingly, asymmetric cell

divisions followed by apoptosis of the cell that is not destined tobecome an md neuron, as seen in the p6 lineage (Fig. 7B), havebeen previously reported in the md-solo lineage in the embryo(Orgogozo et al., 2002). In this case, the situation is more extreme,as the single md neuron is produced by two consecutive asym-metric cell divisions, each of which is accompanied by apoptosis.

Thus, the functions of Poxn in the cells of the p-es organlineages occur at several levels. First, Poxn is necessary to specifythe developmental potentials of the secondary precursors, pIIa andpIIb. In its absence, the developmental potential of pIIa is reducedat the expense of that of pIIb, such that the lineage is transformedto an m-es lineage. Second, Poxn is required to promote mitosisand inhibit differentiation in pIIIa and pIIIb2. Finally, Poxn pre-sumably serves as a differentiation factor in the bipolar neuronthat continues to express it in the kölbchen lineage.

However, Poxn is not required for the differentiation of thethree support cells because these cells are also produced in theabsence of Poxn, albeit with a different lineage. We further assumethat, like in the m-es lineage, the specification of different cellularfates depends on asymmetric cell divisions mediated byN-signaling (Posakony, 1994).

Little is known about the biological functions of the larval p-esorgans. It has been suggested that kölbchen are involved inchemosensory detection. For example, the Gal4 transgene that isactivated by the enhancer of one of the candidate gustatoryreceptors in Drosophila, Gr2B1, has been shown to be expressedin one of the bipolar neurons that innervates the ventral kölbchen,which is consistent with the idea that kölbchen play a role in larvaltaste perception (Scott et al., 2001). The late expression of Poxn ina single neuron might lead to a cell-specific expression of somechemosensory receptors, which allow this neuron to respond tocertain environmental stimuli and process sensory information.

Is the lineage of adult chemosensory bristles similar to that of larvalp-es organs?

Poxn also plays an essential role during development of thechemosensory bristles on legs, wings, and labellum of adultDrosophila (Awasaki and Kimura, 1997; Boll and Noll, 2002;Krstic et al., 2009). Since the lineages of larval and adult m-esorgans are very similar, they might be derived from one ancestrallineage (Lai and Orgogozo, 2004). Similarly, we might argue froman evolutionary point of view that the lineages of poly-innervatedchemosensory bristles in adult flies are similar to those of larvalp-es organs that we describe here. This possibility is furthersuggested by the alterations observed in adult taste bristles ofPoxn null mutants, which are analogous to the changes observedhere for larval p-es organs. Taste bristles on labellum, legs, andanterior wing margins are innervated by one mechanosensory andfour chemosensory neurons (Ray et al., 1993; Lai and Orgogozo,2004). Poxn-Gal4-13 UAS-GFP is expressed in clusters of four(sometimes of three or in only two, observed mainly in thelabellum) chemosensory neurons per taste bristle on the legs,wing margins, and labellum (Boll and Noll, 2002). These Poxn-Gal4-expressing neurons are absent or, in a few cases, reduced toone neuron in Poxn null mutants (Boll and Noll, 2002). The celllineage of taste bristles on the labellum has been analyzedpreviously by labeling replicating cells with BrdU (Ray et al.,1993). However, the proposed model is again similar to the m-eslineage, with a precursor for socket and shaft cells and another forthe sheath cell and all neurons (Ray et al., 1993). Because of thetechnical limitations of BrdU incorporation and since no cell-specific markers were used in this earlier study, the cell lineageof taste bristles on the labellum may also need to be re-examined.

Y. Jiang et al. / Developmental Biology 397 (2015) 162–174 173

Acknowledgments

We are grateful to Gary Struhl, Bruce Hay, Yuh-Nung Jan, theBloomington Drosophila Stock Center, and Developmental StudiesHybridoma Bank for fly stocks and antibodies. We thank AliciaHidalgo for advice on staining of activated Casp-3 and members ofthe Noll lab for discussions and suggestions. We also thankMichael Daube for assistance in the graphical work, and aregrateful to Joy Alcedo for a careful editing of the manuscript. Thiswork was supported by Grant 3100A0-105823 from the SwissNational Science Foundation and the Kanton Zürich.

Appendix A. Supporting information

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.ydbio.2014.10.013.

References

Awasaki, T., Kimura, K.-i., 1997. pox-neuro is required for development of chemo-sensory bristles in Drosophila. J. Neurobiol. 32, 707–721.

Awasaki, T., Kimura, K.-i., 2001. Multiple function of poxn gene in larval PNSdevelopment and in adult appendage formation of Drosophila. Dev. Genes Evol.211, 20–29.

Bellaïche, Y., Schweisguth, F., 2001. Lineage diversity in the Drosophila nervoussystem. Curr. Opin. Genet. Dev. 11, 418–423.

Betschinger, J., Knoblich, J.A., 2004. Dare to be different: asymmetric cell division inDrosophila, C. elegans and vertebrates. Curr. Biol. 14, R674–R685.

Bier, E., Vaessin, H., Shepherd, S., Lee, K., McCall, K., Barbel, S., Ackerman, L.,Carretto, R., Uemura, T., Grell, E., Jan, L.Y., Jan, Y.N., 1989. Searching for patternand mutation in the Drosophila genome with a P-LacZ vector. Genes Dev. 3,1273–1287.

Bodmer, R., Jan, Y.N., 1987. Morphological differentiation of the embryonicperipheral neurons in Drosophila. Roux’s Arch. Dev. Biol. 196, 69–77.

Bodmer, R., Carretto, R., Jan, Y.N., 1989. Neurogenesis of the peripheral nervoussystem in Drosophila embryos: DNA replication patterns and cell lineages.Neuron 3, 21–32.

Boll, W., Noll, M., 2002. The Drosophila Pox neuro gene: control of male courtshipbehavior and fertility as revealed by a complete dissection of all enhancers.Development 129, 5667–5681.

Bopp, D., Jamet, E., Baumgartner, S., Burri, M., Noll, M., 1989. Isolation of two tissue-specific Drosophila paired box genes, Pox meso and Pox neuro. EMBO J. 8,3447–3457.

Brand, A.H., Perrimon, N., 1993. Targeted gene expression as a means of altering cellfates and generating dominant phenotypes. Development 118, 401–415.

Brewster, R., Bodmer, R., 1995. Origin and specification of type II sensory neurons inDrosophila. Development 121, 2923–2936.

Campos-Ortega, J.A., Hartenstein, V., 1997. The Embryonic Development of Droso-phila melanogaster, second ed. Springer-Verlag, Berlin, pp. 177–220.

Dambly-Chaudière, C., Ghysen, A., 1986. The sense organs in the Drosophila larvaand their relation to the embryonic pattern of sensory neurons. Roux’s Arch.Dev. Biol. 195, 222–228.

Dambly-Chaudière, C., Ghysen, A., 1987. Independent subpatterns of sense organsrequire independent genes of the achaete-scute complex in Drosophila larvae.Genes Dev. 1, 297–306.

Dambly-Chaudière, C., Jamet, E., Burri, M., Bopp, D., Basler, K., Hafen, E., Dumont, N.,Spielmann, P., Ghysen, A., Noll, M., 1992. The paired box gene pox neuro: adeterminant of poly-innervated sense organs in Drosophila. Cell 69, 159–172.

Fichelson, P., Gho, M., 2003. The glial cell undergoes apoptosis in the microchaetelineage of Drosophila. Development 130, 123–133.

Fu, W., Noll, M., 1997. The Pax2 homolog sparkling is required for development ofcone and pigment cells in the Drosophila eye. Genes Dev. 11, 2066–2078.

Gho, M., Lecourtois, M., Géraud, G., Posakony, J.W., Schweisguth, F., 1996. Sub-cellular localization of Suppressor of Hairless in Drosophila sense organ cellsduring Notch signalling. Development 122, 1673–1682.

Gho, M., Bellaïche, Y., Schweisguth, F., 1999. Revisiting the Drosophila microchaetelineage: a novel intrinsically asymmetric cell division generates a glial cell.Development 126, 3573–3584.

Ghysen, A., Dambly-Chaudière, C., Aceves, E., Jan, L.-Y., Jan, Y.-N., 1986. Sensoryneurons and peripheral pathways in Drosophila embryos. Roux’s Arch. Dev. Biol.195, 281–289.

Golic, K.G., Lindquist, S., 1989. The FLP recombinase of yeast catalyzes site-specificrecombination in the Drosophila genome. Cell 59, 499–509.

Gutjahr, T., Frei, E., Noll, M., 1993. Complex regulation of early paired expression:initial activation by gap genes and pattern modulation by pair-rule genes.Development 117, 609–623.

Hartenstein, V., 1988. Development of Drosophila larval sensory organs: spatio-temporal pattern of sensory neurones, peripheral axonal pathways and sensilladifferentiation. Development 102, 869–886.

Hartenstein, V., Posakony, J.W., 1989. Development of adult sensilla on the wingand notum of Drosophila melanogaster. Development 107, 389–405.

Hay, B.A., Wolff, T., Rubin, G.M., 1994. Expression of baculovirus P35 prevents celldeath in Drosophila. Development 120, 2121–2129.

Hertweck, H., 1931. Anatomie und Variabilität des Nervensystems und derSinnesorgane von Drosophila melanogaster (Meigen). Z. wiss. Zool. 139,559–663.

Higashijima, S., Michiue, T., Emori, Y., Saigo, K., 1992. Subtype determination ofDrosophila embryonic external sensory organs by redundant homeo box genesBarH1 and BarH2. Genes Dev. 6, 1005–1018.

Hill, A., Boll, W., Ries, C., Warner, L., Osswalt, M., Hill, M., Noll, M., 2010. Origin ofPax and Six gene families in sponges: Single PaxB and Six1/2 orthologs inChalinula loosanoffi. Dev. Biol. 343, 106–123.

Hummel, T., Krukkert, K., Roos, J., Davis, G., Klämbt, C., 2000. Drosophila Futsch/22C10 is a MAP1B-like protein required for dendritic and axonal development.Neuron 26, 357–370.

Jan, Y.N., Jan, L.Y., 1993. The peripheral nervous system. In: Bate, M., Martinez Arias,A. (Eds.), The Development of Drosophila melanogaster, vol. II. Cold SpringHarbor Laboratory Press, New York, NY, pp. 1207–1244.

Jiang, J., Struhl, G., 1995. Protein kinase A and hedgehog signaling in Drosophila limbdevelopment. Cell 80, 563–572.

Krstic, D., Boll, W., Noll, M., 2009. Sensory integration regulating male courtshipbehavior in Drosophila. PLoS One 4, e4457.

Lai, E.C., Orgogozo, V., 2004. A hidden program in Drosophila peripheral neurogen-esis revealed: fundamental principles underlying sensory organ diversity. Dev.Biol. 269, 1–17.

Lee, T, Luo, L., 1999. Mosaic analysis with a repressible cell marker for studies ofgene function in neuronal morphogenesis. Neuron 22, 451–461.

Noll, M., 1993. Evolution and role of Pax genes. Curr. Opin. Genet. Dev. 3, 595–605.Orgogozo, V., Schweisguth, F., 2004. Evolution of the larval peripheral nervous

system in Drosophila species has involved a change in sensory cell lineage. Dev.Genes Evol. 214, 442–452.

Orgogozo, V., Schweisguth, F., Bellaïche, Y., 2001. Lineage, cell polarity and inscute-able function in the peripheral nervous system of the Drosophila embryo.Development 128, 631–643.

Orgogozo, V., Schweisguth, F., Bellaïche, Y., 2002. Binary cell death decisionregulated by unequal partitioning of Numb at mitosis. Development 129,4677–4684.

Posakony, J.W., 1994. Nature versus nurture: asymmetric cell divisions in Drosophilabristle development. Cell 76, 415–418.

Ray, K., Hartenstein, V., Rodrigues, V., 1993. Development of the taste bristles on thelabellum of Drosophila melanogaster. Dev. Biol. 155, 26–37.

Reddy, G.V., Rodrigues, V., 1999. A glial cell arises from an additional divisionwithinthe mechanosensory lineage during development of the microchaete on theDrosophila notum. Development 126, 4617–4622.

Robinow, S, White, K., 1991. Characterization and spatial distribution of the ELAVprotein during Drosophila melanogaster development. J. Neurobiol. 22, 443–461.

Scott, K., Brady Jr., R., Cravchik, A., Morozov, P., Rzhetsky, A., Zuker, C., Axel, R., 2001.A chemosensory gene family encoding candidate gustatory and olfactoryreceptors in Drosophila. Cell 104, 661–673.

Struhl, G., Basler, K., 1993. Organization of Wingless protein in Drosophila. Cell 72,527–540.

Vaessin, H., Grell, E., Wolff, E., Bier, E., Jan, L.Y., Jan, Y.N., 1991. prospero is expressedin neuronal precursors and encodes a nuclear protein that is involved in thecontrol of axonal outgrowth in Drosophila. Cell 67, 941–953.

Yu, S.-Y., Yoo, S.J., Yang, L., Zapata, C., Srinivasan, A., Hay, B.A., Baker, N.E., 2002. Apathway of signals regulating effector and initiator caspases in the developingDrosophila eye. Development 129, 3269–3278.

Y. Jiang et al. / Developmental Biology 397 (2015) 162–174174

clone types

hours AEL all cellsthree

supportcells

fourneurons neurons neuron

socketcell

shaft andsheath

cell bipolar

bipolarand md others

2 – 3

4 – 55 – 6

3 – 483 7 10 0 0 0 0 0

42 24 31 0 0 0 0 07 44 53 2 3 4 1 10 10 13 10 41 33 30 6

Table S1

Table S1. Numbers of clones, marked after various time intervals by a 30-min heat shock at 32°C in the dorsal and lateral kölbchen lineages of y w hs-flp (homozygous or over Y); UAS>CD2 y+>CD8::GFP (homozygous or over CyO); Poxn-Gal4ups1f (homozygous or over TM6B) embryos. Clone types are indicated at the top and time intervals of heat shock on the left. Time intervals correspond to development at 25°C, but development before and after heat shock was at 18°C. The clones counted as ‘others’ are double clones induced in two different precursors of the lineage. No single-cell clones other than socket-cell clones were observed, presumably because these cells originated only after the heat-shock intervals used.

two

md

94 10 16 0 0 0 0 0 056 23 28 0 0 0 0 0 0

4 33 32 2 1 0 0 0 00 n.d.

n.d.8 33 35 1 8 2 0

0 0 8 6 16 49 36 15

hours AEL

clone types

Table S2

p6 cellsp6 cells

eg cellseg cellsand support

cellsneurons socket

cell

shaft andsheath

cell neuronsbipolar

neuron

2 – 3

4 – 55 – 66 – 7

3 – 4

Table S2. Numbers of clones induced after various time intervals by a 30-min heat shock at 32°C in the p6 and epidermal gland (eg) lineage of embryos. Clone types are indicated at the top and time intervals of heat shock on the left. Time intervals correspond to development at 25°C, but development before and after heat shock was at 18°C. n.d., not determined. Genotypes of embryos are y w hs-flp (homozygous or over Y); UAS>CD2 y+>CD8::GFP (homozygous or over CyO); Poxn-Gal4ups1f (homozygous or over TM6B) and, in addition, y w hs-flp (homozygous or over Y); E7-2-36 UAS>CD2 y+>CD8::GFP (homozygous or over CyO); Poxn-Gal4ups1f (homozygous or over TM6B) for clones induced after intervals of 5-6 and 6-7 hours AEL.

37 1111

15 0 0 0 020 28 0 0 0 0

2 12 13 6 4 9 9

clone types

Table S3

hours AEL all cellssheathcell andneurons

socketcell

shaftcell

4 – 55 – 6

3 – 4

socketand shaft

cell

sheath celland bipolar

neuron

md neuron ormd and bipolar

neuron

Table S3. Numbers of clones marked after various time intervals by a 30-min heat shock at 32°C in the dorsal/lateral and ventral kölbchen lineage of Poxn null mutant embryos. Clone types are indicated at the top and time intervals of heat shock on the left. Time intervals correspond to development at 25°C, but development before and after heat shock was at 18°C. Genotypes of embryos are y w hs-flp (homozygous or over Y); Poxn∆M22 UAS>CD2 y+>CD8::GFP; Poxn-Gal4ups1f (homozygous or over TM6B).

Fig. S1. Clones of ventral kölbchen. (A-D) Clones of stage 16 embryos are shown that were generated at 3-4, 4-5, 5-6, or 6-7 hours AEL. Embryos were pooled and immunostained with antibodies directed against the proteins indicated in the corresponding colors below each panel. Genotypes of embryos are as described in the legend to Fig. 2. For a detailed description of clones, see text.

Fig. S2. Clones of hair 3 in the presence and absence of Poxn protein. (A-L) Clones of stage 16 embryos are shown that were generated at 3-4, 4-5, 5-6, or 6-7 hours AEL in the presence of wild-type Poxn protein (A-G) or in its absence (H-L). Embryos were pooled and immunostained with antibodies directed against the proteins indicated in the corresponding colors below each panel. The overlap of β-gal (red) with Elav (blue) is purple (D). Note that in (G) the more ventral bipolar neuron weakly expresses BarH1 (red), which is also expressed in the sheath cell of h3, adjacent and to the right of the clone (not labeled by GFP). There are several md neurons visible in (G), the most ventral of which, because of its position (D), is probably the one that belongs to h3. Genotypes of embryos are as described in the legend to Fig. 2 (in the presence of Poxn) and Fig. 5 (in the absence of Poxn). For a detailed description of clones, see text.

Fig. S3. No apoptosis in the wild-type lineage of hair 3. (A-D) None of the GFP-labeled cells of the wild-type h3 lineage stain for activated Casp-3 in stage 12 (A), stage 13 (B), stage 14 (C), and stage 15 (D) y w; UAS-CD8::GFP; Poxn-Gal4ups1f (homozygous or over TM3) embryos. Note that in panel (A) the lineage shows only five cells, probably because the largest cell is still in the process of division.

Fig. S4. No apoptosis in the wild-type lineage of kölbchen. (A, B) None of the GFP-labeled kölbchen cells stain for activated Casp-3 in stage 13 (A) and stage 14 (B) y w; UAS-CD8::GFP; Poxn-Gal4ups1f (homozygous or over TM3) embryos. Asterisks indicate apoptotic cells in neighboring tissues that express activated Casp-3. (C) No supernumerary cells are observed among the CD2-labeled kölbchen cells when apoptosis is inhibited by the expression of P35 in a y w hs-flp/(w or Y); UAS>CD2 y+>CD8::GFP/UAS-p35; Poxn-Gal4ups1f/+ embryo at stage 16.

Fig. S5. No apoptosis in the kölbchen lineage of Poxn null mutants. (A, B) None of the CD-2-labeled kölbchen cells stain for activated Casp-3 in stage 13 (A) and stage 14 (B) y w hs-flp; Poxn∆M22 UAS>CD2 y+>CD8::GFP; Poxn-Gal4ups1f (homozygous or over TM6B) embryos. Asterisks indicate apoptotic cells in neighboring tissues that express activated Casp-3. (C, D) No supernumerary cells are observed among the CD2-labeled kölbchen cells when apoptosis is inhibited by the expression of P35 in y w hs-flp; Poxn∆M22 UAS>CD2 y+>CD8::GFP/Poxn∆M22 UAS-p35; Poxn-Gal4ups1f (homozygous or over TM6B) embryos. The ratio of the number of lineages producing 5 (18; C) or 6 (8; D) cells remains the same as observed in embryos without P35 expression (Fig. 5A and B).